Multistep Prediction Analysis of Pure Pursuit Method for Automated Guided Vehicles in Aircraft Industry

Abstract

The pure pursuit (PP) method has been widely employed in automated guided vehicles (AGVs) to address path tracking challenges. However, the traditional pure pursuit method exhibits certain limitations in tracking performance. For instance, selecting a look-ahead point that is too close can lead to oscillations during tracking, while selecting one that is too far away can result in slow tracking and corner-cutting issues. To address these challenges, this paper proposes a multistep prediction pure pursuit method. First, the look-ahead distance calculation equation is adjusted by incorporating path curvature, allowing it to adaptively adjust according to road conditions. Next, to avoid oscillations caused by constant changes in the look-ahead distance, this paper adopts the prediction concept of model predictive control (MPC) to make multistep predictions for the pure pursuit method. The final input is derived from a linear weighted combination of the multistep prediction results. Simulation analyses and experiments demonstrate that the multistep predictive pure pursuit method significantly enhances the tracking performance of the traditional pure pursuit method.